Merge pull request #964 from KnowledgeCaptureAndDiscovery/dev

Dev

Author: dgarijo

README.md

@@ -27,13 +27,12 @@ Given a readme file (or a GitHub/Gitlab repository) SOMEF will extract the follo
 - **Build file**: Build file(s) of the project. For example, files used to create a Docker image for the target software, package files, etc.
 - **Citation**: Preferred citation(s) as the authors have stated in their readme file. SOMEF recognizes Bibtex, Citation File Format files and other means by which authors cite their papers (e.g., by in-text citation). 
 For CITATION.cff files, SOMEF now generates two separate entries: one for the software tool and another for the preferred citation (if available). This ensures metadata like DOI or version is correctly assigned to each entity.
-We aim to recognize the following properties:
+We recognize the following properties:
   - Title: Title of the publication
   - Author: list of author names in the publication
   - URL: URL of the publication 
   - DOI: Digital object identifier of the publication
   - Date published
-  - Version: Software version (if applicable)
   - Journal: Journal name where the paper was published
   - Year: Year of publication
   - Pages: Page range in the journal

docs/citationcff.md

@@ -18,7 +18,7 @@ These fields are defined in the [CITATION.cff specification](https://citation-fi
 | license - value | license[i].result.value | license |
 | license - spdx_id | license[i].result.spdx_id | license |
 | license - name | license[i].result.name | license  |
-
+| version - value | version[i].result.value	|version
 ---
 
 *(1)* 

docs/index.md

@@ -34,7 +34,7 @@ Given a readme file (or a GitHub repository) SOMEF will extract the following ca
 - **Build file**: Build file(s) of the project. For example, files used to create a Docker image for the target software, package files, etc.
 - **Citation**: Preferred citation(s) as the authors have stated in their readme file. SOMEF recognizes Bibtex, Citation File Format files and other means by which authors cite their papers (e.g., by in-text citation). 
 For CITATION.cff files, SOMEF now generates two separate entries: one for the software tool and another for the preferred citation (if available). This ensures metadata like DOI or version is correctly assigned to each entity.
-We aim to recognize the following properties:
+We recognize the following properties:
   - Title: Title of the publication
   - Author: list of author names in the publication
   - URL: URL of the publication 

src/somef/process_files.py

@@ -576,12 +576,20 @@ def get_file_content_or_link(repo_type, file_path, owner, repo_name, repo_defaul
 
                 if yaml_content:
                     license_value = yaml_content.get("license")
+                    version_value = yaml_content.get("version")
+
                     logging.info(f"Extracted license value from CFF: {license_value}")
                     if license_value:
                         if isinstance(license_value, list):
                             license_value = license_value[0]
                         parse_license_cff(license_value, metadata_result, url)
 
+                    logging.info(f"Extracted version value from CFF: {version_value}")
+                    if version_value:
+                        if isinstance(version_value, list):
+                            version_value = version_value[0]
+                        parse_version_cff(version_value, metadata_result, url)
+
                     root_result = parse_cff_root(yaml_content, metadata_result,url)
                     root_result[constants.PROP_VALUE] = file_text
                     # root_result[constants.PROP_TYPE] = constants.FILE_DUMP
@@ -722,7 +730,7 @@ def parse_cff_root(yaml_content, metadata_result, url):
 
     result[constants.PROP_TITLE] = yaml_content.get("title")
     result["authors"] = parse_authors_citation(yaml_content.get("authors", []))
-    result[constants.PROP_VERSION] = yaml_content.get("version")
+    # result[constants.PROP_VERSION] = yaml_content.get("version")
     result[constants.PROP_DOI] = yaml_content.get("doi")
     result[constants.PROP_URL] = yaml_content.get("url")
     result[constants.PROP_TYPE] = constants.SOFTWARE_APPLICATION
@@ -789,4 +797,25 @@ def parse_license_cff(license_value, metadata_result, url):
         logging.error(f"Error parsing license from CFF: {str(e)}")
 
 
+def parse_version_cff(version_value, metadata_result, url):
+    """
+    Parses the version from a CFF file and adds it to the global version metadata.
+    """
+    try:
+
+        version_result = {
+            constants.PROP_VALUE: str(version_value),
+            constants.PROP_TYPE: "String" 
+            
+        }
+
+        metadata_result.add_result(
+            constants.CAT_VERSION,
+            version_result,
+            1, 
+            constants.TECHNIQUE_FILE_EXPLORATION,
+            url
+        )
 
+    except Exception as e:
+        logging.error(f"Error parsing version from CFF: {str(e)}")

src/somef/test/test_JSON_export.py

@@ -630,15 +630,25 @@ def test_new_properties_citation_issue_935(self):
         assert software_entry is not None, "Software citation (root) not found"
         sw_result = software_entry["result"]
         assert sw_result["title"] == 'SOMEF: Software metadata extraction framework'
-        assert sw_result["version"] == "0.1.0"
+        # assert sw_result["version"] == "0.1.0"
         assert "doi" not in sw_result or sw_result.get("doi") is None # it is in preferred (referencePublication) but not in the root
 
         assert preferred_entry is not None, "Preferred citation (article) not found"
         pref_result = preferred_entry["result"]
         assert pref_result["title"] == "A Framework for Creating Knowledge Graphs of Scientific Software Metadata"
         assert pref_result["doi"] == "10.1162/qss_a_00167"
         assert pref_result["journal"] == "Quantitative Science Studies"
-        assert "version" not in pref_result # it is in the root in citation but not in the preferred (referencePublication)
+        # assert "version" not in pref_result # it is in the root in citation but not in the preferred (referencePublication)
+
+        versions = json_content.get(constants.CAT_VERSION, [])
+        cff_version_entry = next(
+            (v for v in versions if "CITATION.cff" in v.get("source", "")),
+            None
+        )
+
+        # 2. Validamos que la versión existe en su nueva ubicación
+        assert cff_version_entry is not None, "Version from CFF not found in global version field"
+        assert cff_version_entry["result"]["value"] == "0.1.0"
 
         os.remove(test_data_path + "test_new_properties_citation_issue_935.json")